1. Introduction
Autism is a neurodevelopmental profile (or ‘neurotype’) traditionally characterised in part by social communication deficits (APA 1994; APA 2013; WHO 1994). More recent iterations of both diagnostic manuals are beginning to grapple with how social and environmental context affects autistic people’s behaviour. For example, the DSM-5-TR and the ICD-11 both note that autistic ‘symptoms’ may not be visible if autistic people are in appropriate environments, or they have learned to mask (APA 2022; WHO 2022). Nonetheless, both manuals still refer to ‘persistent deficits in social communication’ as one of the two defining features of autism (APA 2022; WHO 2022).
Recent theoretical work on autism has suggested this framing may not be correct, however. The double empathy problem, in particular, claims that autistic social communication is not deficient per se (Milton 2012). Rather, social communication dysfunction between autistic and non-autistic people is relational and mutual. It occurs due to the difficulties experienced by people of different neurotypes in understanding and relating to one another’s lived experiences and perceptions of the world (Milton, Waldock and Keates 2023; Milton 2012; Milton, Heasman and Sheppard 2018).
Recent experimental research supports this framing of autistic social communication. Autistic people transfer information between one another as effectively as non-autistic people, with difficulties only occurring when information transfer happens between autistic and non-autistic people (Crompton, Ropar et al. 2020). Interactions between autistic dyads were given rapport ratings by participants and observers equivalent to those given to non-autistic dyads, with difficulties again arising only in a cross-neurotype context (Crompton, Sharp et al. 2020; Jones et al. 2023), despite traditional, non-autistic markers of rapport, like backchannelling and eye contact, being reduced in autistic-autistic social interactions (Rifai et al. 2022). And, while autistic people may struggle to understand non-autistic people’s mental states and social cues, the reverse is also true: non-autistic people also struggle to interpret autistic mental states and social cues (Alkhaldi, Sheppard and Mitchell 2019; Casartelli et al. 2020; Edey et al. 2016; Heasman 2018; Sheppard et al. 2016).
The double empathy problem also suggests that autistic people may find social interactions and communication with other autistic people more successful and enjoyable than with non-autistic people (Milton 2012). This, too, seems to be borne out by research. Several qualitative studies have also found that autistic people report preferring the company of other autistic people to that of non-autistic people (Bolis et al. 2020; Crompton, Hallett et al. 2020; Morrison et al. 2020; Watts et al. 2024). And, while autistic observers can detect ‘problematic’ social behaviours and social awkwardness in other autistic people, these do not influence their social interest in the same way they do for non-autistic observers (DeBrabander et al. 2019; Sng et al. 2020). In summary, there is a growing evidence base that differences in social communicative features between autistic and non-autistic people are a consequence of an autistic interactional ‘style’ – and the ways in which this differs from the non-autistic interactional style – rather than a deficit (Bolis et al. 2020; Heasman and Gillespie 2019; Rifai et al. 2022).
These differences in style between autistic and non-autistic interlocuters seem to be primarily driven by nonverbal or paralinguistic phenomena, rather than by the actual content of the conversation (Grossman 2015; Sasson et al. 2017). This includes acoustic and prosodic properties of speech (Bone et al. 2015) as well as differences in body language, movement, and facial expressions (Casartelli et al. 2020; Zane et al. 2018). At a more abstract level, studies have found differences in features such as ‘flow’ and ‘tuning-in’ (Williams, Wharton and Jagoe 2021) in autistic-autistic social interactions, and distinct autistic interactional features such as ‘generous assumption of common ground’ and ‘low demand for coordination’ (Heasman and Gillespie 2019).
Better understanding of how autistic and non-autistic interactional styles differ could help uncover the underpinnings of the double empathy problem and provide pathways for bridging the double empathy gap that currently exists between these two groups. However, much of our current data on autistic social and communicative behaviour comes from non-autistic researchers, working under a deficit model, without appropriate autistic lived experience input. This research treats non-autistic social behaviour as the default, and the standard against which autistic behaviour is measured; where autistic behaviour deviates from this standard, it is considered ‘deficient’. This leads to potential misinterpretation or misattribution of autistic social and communicative behaviours due to the non-autistic researcher’s unexamined biases (Bottema-Beutel et al. 2023; Dickter et al. 2020; Duffy and Dorner 2011).
Our study aimed to rectify this by exploring autistic accounts of experiences of face-to-face social communication and interaction, using autistic expertise and lived experience drawn from both the participants and an autistic member of the research team. By collecting primary data about autistic experiences, and interpreting these through an autistic analytic lens, we aimed to uncover features of the autistic social communicative style, and possible loci of disjunctions of communicative style between autistic and non-autistic people, in order to better understand the origins of the double empathy gap.
We took the neurodiversity paradigm (Dwyer 2022; Walker 2021) and the social model of disability (Barnes 2019) as theories for this study’s research paradigm, and these informed our development of our guiding interest and research questions. Our guiding interest (O’Reilly 2011) was exploring autistic people’s feelings and reflections on preferred communication styles and preference – in the context of interactions with both with other autistic people, and with non-autistic people. We were interested primarily in specific paralinguistic features of communication (e.g. body language and gestures, turn-taking, tone of voice) and markers of rapport. Emerging from our guiding interest, we developed research questions to help shape the boundaries of our data collection and inform our analysis:
Do autistic people feel that paralinguistic features of verbal communication commonly used by non-autistic people are important to, or necessary for, successful communication?
Do autistic people feel there is a difference between how they communicate face-to-face with other autistic people, versus how they communicate with non-autistic people? If yes, what are those differences?
What are other aspects of autistic communication not covered by the above questions that autistic people feel a) researchers should know about or b) researchers should be asking questions about?
Given the qualitative and exploratory nature of the study, we had no specific hypotheses regarding what the answers to these research questions might be. However, as mentioned above, we were at the outset of the study assuming that we would find specific paralinguistic/nonverbal ‘features’ of autistic communication.
2. Methods
Ethical approval for this study was given by the University of Edinburgh Medical Research Ethics committee. This paper uses O’Brien et al.’s (2014) standards for reporting qualitative research as a guide to improve clarity and transparency of the research and analysis process.
2.1 Preregistration
This study was preregistered at https://osf.io/9cgz7. However, several substantiative elements of the current study differ from the preregistration, including:
a shift from a planned focus on exploring an autistic social communicative ‘style’ in terms of specific behaviours or markers of rapport. Data generated by participants focused on social communication ‘values’ and experiences, rather than generating specific markers, and this resulted in a shift in focus for the analysis and discussion
change in number of participants in the study, due to participant drop-out during the study
a limitation of the member-checking process due to participant non-responsiveness (see below)
critical realist thematic analysis (Fryer 2022) being used in place of framework analysis (Srivastava and Thomson 2009). This was due to the need for a more inductive approach to the data than we had anticipated when the focus group did not elicit any paralinguistic/nonverbal markers of rapport and instead generated richer and more generalised descriptions of autistic and social communicative style.
2.2 Researcher positionality and reflexivity
The first author (who conducted the focus group and analysis) is autistic and regularly talks to other autistic people about their lived experience of being autistic. She comes from a linguistic academic background. The other members of the research team are not autistic. Collectively, the research team’s position is that that autistic social communication is not inherently and universally deficient; and that autistic traits, behaviours, and cognition should not be considered disordered simply because they differ from the neuromajority. We use frameworks such as the neurodiversity paradigm (Dwyer 2022; Walker 2021), the social model of disability (Barnes 2019), and the double empathy problem (Milton 2012) when theorising about and researching autism.
2.3 Participants
Participants were required to be a) adults, who b) were fluent in English, c) had grown up in the UK, and d) had received a clinical diagnosis of autism. We excluded people who had not grown up in the UK to attempt to minimise the impact of cultural differences on participant accounts of social interaction and communication. We only included participants with a clinical diagnosis because, initially, we were interested in looking for specific features of autistic behaviour and rapport, and wanted (given the small number of participants, which prevented other common methods for including participants with self-diagnosis such as e.g. comparing the findings from the overall group of participants to the findings from a ‘formal diagnosis only’ subgroup as a means of validation) to be sure all participants were indeed autistic. Whilst there is no data on the accuracy of self-diagnosis, some researchers have expressed concerns that people may self-diagnose as autistic when another similar but distinct diagnosis would be a better fit, because autism is more well-known than many other diagnoses (Fellowes 2023).
Participants were recruited via the Patrick Wild Centre participant database and via a Twitter advert. Informed consent was given to take part in the study. Potential participants provided contact details, demographic information, and consent as part of an initial screening process. Twenty four potential participants provided their details; two were excluded due to not meeting the study inclusion criteria. Of the remaining 22 potential participants, we used maximum variation sampling (Palinkas et al. 2015) to select 10 focus group participants. One participant did not comment at all during the focus group.
We therefore had a total of nine participants who contributed data to the analysis. Based on recommendations by Allen and Wiles (2016), participants selected their own pseudonym (which did not contain any identifying information such as nicknames). The contributing participants were Zel, AppleFrog, Pumpkin, The Laird, Barney, Owain, Bert, Lily, and Alexander. Demographic data for participants are shown in Table 1. Participants were compensated with a £20 online shopping voucher of their choice for taking part, which was sent to them upon close of the focus group.
Table 1
Demographic details of participants who contributed to the focus group.
| Gender | Women = 4 Non-binary = 1 Man = 4 |
| Age (yrs) | 45.2 ± 14.8 (mean ± SD) 23–70 (range) |
| Ethnicity | White = 8 Mixed race/other ethnicity = 1 |
| Index of Multiple Deprivations (IMD) decile* | 4.2 ± 2.8 (mean ± SD) 1–9 (range) |
| Additional conditions** | ADHD (5) Dyslexia (1) Dyspraxia (1) Dyscalculia (1) |
[i] *This data was available for only six of the nine participants.
**These are total numbers of participants with each condition, per condition. Some participants may have had more than one additional condition; these have been listed separately, to prevent risk to participant anonymity.
2.4 Data collection
We collected data using an online focus group. This was run by the first author via the website Quallie (2024). Quallie provides a GDPR-compliant, forum-style experience, where the researcher can set up a focus group project with pages for different topics. Participants received a unique log-in to the focus group project, and could then comment, reply to other participants’ comments, or ‘like’ those comments.
The focus group was conducted online 1) to limit wellbeing risks to participants (both due to the Covid-19 pandemic during data collection in 2021, and due to potential anxiety or stress caused by a novel location/situation), and 2) to facilitate UK-wide recruitment. It also kept participants more anonymous from one another than in-person or video-conferencing data collection would have. Given the personal nature of the data we were soliciting from participants, and the fact that some participants may not have been ‘out’ as autistic publicly, we hoped that this additional anonymity would encourage greater participant engagement.
There were initially four topics available to participants. These are detailed in Table 2 below. On the suggestion of a participant (Pumpkin), we added a fifth topic page, also detailed. There was also a page with an introduction to the study, which included guidance on participant conduct during the group. See Appendix A in the supplemental material available at https://osf.io/9cgz7 for the full text of all the pages shown to the participants.
Table 2
Focus group topics.
| TOPIC NUMBER | TOPIC NAME |
|---|---|
| 1) | Signalling in conversations (how participants indicated they liked/disliked/were bored with etc. someone during conversations) |
| 2) | Natural communication habits (what sort of things participants and others did during interactions that felt natural or unnatural) |
| 3) | Talking to other autistic people (what the differences between interacting with autistic vs. non-autistic people were) |
| 4) | Anything else (any other important information about autistic social interaction or communication that participants felt we should know, but weren’t covered by the other topics) |
| 5) | Positive social interactions* (what participants felt made a social interaction positive for them) |
The focus group was open for two weeks. Participants were sent an email upon opening and close of the group, and three emails between those times encouraging them to interact with the focus group. The first author checked the focus group at minimum once a day, to respond to participants’ comments and to post new prompts under each topic if needed.
2.5 Data analysis
HEAS led the analysis. During data collection, HEAS took notes on their initial impressions of the data (as per Braun and Clarke 2006) to monitor the influence of her positionality and to suggest initial codes for the analysis. Data were analysed in NVivo (Lumivero 2020) with critical realist thematic analysis (Fryer 2022), supplemented with additional analytic steps drawn from Flowers et al. (2016) and Naeem et al. (2023). Critical realist thematic analysis has many similarities to Braun and Clarke’s (2006) better-known reflexive thematic analysis, including many of the same analytic steps. Indeed, Fryer describes critical realist thematic analysis as an attempt to build upon Braun and Clarke’s model ‘in a way that brings back ontological realism’ (Fryer 2022), such that thematic analysis can produce not just stories about people’s experiences but also causal explanations.
The data were initially sorted into semantic-level, descriptive, and data-led codes, developed inductively by HEAS. Codes were then inductively organised into themes and subthemes. Using these newly developed themes and subthemes, the dataset was then re-analysed. HEAS re-read the entirety of the data and, having removed the codes made during the initial round of analysis, re-assigned segments of data into each of the themes from the first round of analysis. After this second round of analysis, the themes were revisited. By closely reading the data segments associated with each theme – as well as data that did not fit into any of the extant themes – the themes were revised, refined, and re-organised (See Figure 1). A third and final round of analysis was then done using these revised themes. This again involved re-reading and re-coding the entirety of the data, using the revised themes and subthemes.
Figure 1
The five themes developed from the thematic analysis.
Attempts at member-checking the analysis (see Appendix B in the supplemental material available at https://osf.io/9cgz7) were unsuccessful; only two participants responded, both of whom left positive but uninformative feedback. However, given the unclear utility of member checking in general (Morse 2015), and the specific objections to it in a thematic analysis context (Braun and Clarke 2022), we do not consider this detrimental to the overall analysis.
3. Thematic Analysis
Theme 1: Autistic experiences of self, attention, and environment
Participants’ fundamental experiences of social communication and social interactions were, they felt, distinct from non-autistic peoples’ experiences in several ways.
1a. Limited social battery
Conversations and social interactions were often very tiring for participants. Participants struggled to pay attention during conversations and often found themselves ‘zoning out’: ‘[Having a conversation] requires effort, presence, and energy. They say pick your battles, and metaphorically I do this with conversations with all types of people. Sometimes it feels like a battle to stay focused and not zone out.’1 (The Laird). This was especially true if the conversation was not particularly interesting or enjoyable: ‘If I am not enjoying the conversation then I find it a real effort to keep it going, though I can still do it, but I get tired very quickly and will become quickly irritated.’ (Barney)
1b. Fragmented attention
Participants could also struggle to keep track of what was happening in conversations and would sometimes forget important details from a conversation afterwards. This was a particular issue when (a) others went off on tangents or interrupted the participants during the conversation: ‘[I]f I am interrupted I have to go back to page 1 and start again. I just lose my flow.’ (The Laird), or (b) sensory demands were high (primarily related to auditory processing, though heat, light, smell, and clothing texture were also reported):
I find if there is a lot of background noise, I tend to focus on that rather than what a person is saying to me. I also find that I can’t hear what the other person is saying, like being partially deaf, my ears and brain just can’t seem to work out what the other person is saying and it tends to come through as one long mumbled sentence. (Pumpkin)
1c. Issues self-monitoring and self-managing
However, participants themselves often went off on tangents or interrupted others during conversations: ‘When I’m talking to my girlfriend about something that interests me, and she is speaking, sometimes I feel a great urge to interrupt. […] I’ll interject and then say “Sorry, I interrupted you”.’ (Alexander). They also often found themselves monologuing and could forget to let others talk: ‘I watch myself for talking for too long more carefully around non-autistic people. Close friends and autistics are more tolerant of lengthier and more passionate talk about my special interests.’ (Zel). These were particularly common occurrences when participants were talking about special interests of theirs.
1d. Nothing feels natural
Despite this, participants did not describe these behaviours as being part of a natural social communicative style. Participants were unsure of whether any social communication behaviours they had were natural: ‘I am really not sure any of my conversation skills are natural. It all feels thought through and performed.’ (Barney). They reported that they often felt they had ‘picked up’ behaviours from other people either consciously or unconsciously: ‘I pick up accents and rhythms of speech quite easily. My accent changed a lot when I was at uni. I also had a pretty bad stammer in my pre-teens, but I didn’t stammer half as much using somebody else’s voice.’ (Alexander).
Theme 2: Autistic expectations about how social interactions should work
Participants had distinctive ideas about how social interactions should work, and what constituted a good or enjoyable social interaction – specific ‘social communicative values’ – that they felt differed from non-autistic people’s:
There is a difference of typical values that determines what happens in a conversation. NDs [neurodivergent people] seem more preoccupied with information, accuracy, solutions, truth, correctness, substance. NTs [neurotypical people] seem more preoccupied with status, power, appearance, gathering allies, defining groups. […] I’m just wondering if these underlying values determines what kinds of things typically happen in conversations. (Bert)
These values were also influenced by the differing social communication experiences that participants had reported. Three factors had a particularly significant impact: the purpose of the interaction, the honesty and authenticity of the interlocutor, and feelings of mutual reciprocity and effort.
2a. Purpose
Social interactions with a specific purpose – such as completing a task, cooperating on an activity, or sharing information – were easier to navigate than other types of social interactions. Participants felt they were often quite good at these types of interactions:
I could have really good conversations with colleagues about work, especially technical matters. I was known as a very clear communicator and respected for being able to explain things clearly and unambiguously. […] But ask me to have an informal chat in a team meeting with the same colleagues and I would be at a total loss and usually say something ‘rude’ or inappropriate. (Barney).
Interactions where the purpose was discussing something one (or both) of the conversation participants was passionate about were particularly desirable: ‘I’ve also been really interested to hear other people talk about something they are passionate about, even if I have no knowledge or prior interest in that topic.’ (Zel) Conversations about topics of special interest to participants were less tiring than other types of interaction, and participants found it easier to pay attention during them: ‘I often find it changes the context and environment of an ‘encounter’ if I am talking about something that I am passionate about.’ (Lily)
However, participants could get so carried away during these conversations that it became difficult to remember to let others speak (as described in Theme 1) and perform masking behaviours (such as the ones described in Theme 5):
[W]hen I start talking about my passion I go to another world as I recall facts, give evidence and talk about the subject. […] It’s in these moments that [an autistic person’s] true self shines through, because they become [so] consumed by what they are talking about that they forget to do all the things like masking or overthinking and analysing. They are so caught up in what they are talking about, that they become the truest version of themselves, no holds barred. (Pumpkin)
2b. Honesty and authenticity
Being honest, accurate, and direct when communicating was very important for participants, and they found it frustrating when other people failed to do this:
Failing to answer questions directly and not being honest are my two biggest problems with NTs. […] I find this makes things worse as I can’t figure out what they are really trying to say. I understand I am supposed to read between the lines but that rarely ends well! (Barney)
However, they also acknowledged that other people could find their own directness and honesty rude or upsetting, and that this could cause issues when interacting with others (especially non-autistic people):
Usually it’s OK [to be blunt with other neurodivergent people], sometimes it causes offence. I once told an ADHD person they had been boring a guest of ours. The ensuing row lasted 24 hours. In the end she thanked me for telling her. (Bert)
Being authentic as a person was also highly valued. Many participants felt that being inauthentic during interactions – even for the sake of politeness – was a form of deception, and therefore negative: ‘[S]ometimes I feel like I’ve tricked somebody into having a conversation by acting nice or interested when I’m really not. “That person thinks I’m a good person, but really I’m just pretending to be one”.’ (Alexander)
2c. Mutual respect and effort
Being respectful towards others, and putting effort into social interactions, was very important for participants:
I find it difficult if the other person is distracted or not paying full attention during a conversation, which I suspect is because I have to make so much effort to have a conversation that it seems rude that they are not trying as hard as me. (Barney)
This respect involved acknowledging that every person and relationship was unique. Participants felt it was necessary to learn each new person’s communication style:
I think with different autistic people I develop different systems of communication that work for our dynamic. Once developed, they tend to stay the same, or change slowly. This structure and predictability allows for creativity and variance within that. It’s the opposite of restrictive, paradoxically.’ (Bert)
This involved actively thinking about other people’s perspectives and needs, and then accommodating them by either modifying your own behaviour or accepting unusual behaviours of theirs:
[others say autistic people lack] empathy when in fact we are the most empathetic people because we have to work so hard to understand other that we have learnt to see things from many different point of view, including those that don’t necessarily align with our stance on a subject. (Pumpkin)
Participants themselves often had strong boundaries around how they were willing to socially interact with others, which they expected people to respect. These boundaries often related to a desire for routine and certainty regarding social interactions, which often made interactions easier for participants to engage with:
I like to know what the plan is, exactly. Where we will meet, what time, how long we will be out for, what we might want to talk about etc. The more I can plan and prepare the more comfortable I will be that it will go well. (Barney)
Theme 3: Conflicts between autistic and neurotypical-normative expectations
Given the importance of respect, effort, and understanding, participants found it upsetting that they often felt they were not respected, understood, or accommodated by neurotypical people during social interactions.
Neurotypical people often had fixed, inflexible expectations about how social interactions should work, and could struggle to cope when those expectations were violated: ‘[I]t’s not that [autistic people] do things differently, it’s that you don’t do the things that are expected [by neurotypical people.]’ (Pumpkin). Participants felt that neurotypical people expected them to be able to intuit and abide by these implicit social rules, but participants found them confusing and mysterious:
What NTs [neurotypical people] say and what they mean are often very different and reading between the lines can be essential. Of course, sometimes they do not want you to pick up on what they are not saying and when you do and bring that up they can get very cross. That is where the unwritten social rules kick in and I can fall foul of them. (Barney)
Neurotypical people were also reported as having different ideas about what constituted honesty and clarity during communication, which could cause conflict between participants and others: ‘It sometimes means I run into people who don’t choose their words carefully and so misinterpret [things that I say]. I have had to tell people before ‘I’m not being sarcastic. You just don’t think I mean what I say, but I do.’’ (Alexander).
These factors combined to make interacting with neurotypical people an unpredictable and difficult experience: ‘I tend to treat conversations like I am a politician being interviewed on TV. Every response is thought through and considered in detail. I am expecting to be caught out or tripped up on some fact.’ (Barney). Participants reported that neurotypical social inflexibility, and tendency to socially punish people who failed to follow neurotypical social rules, resulted in powerfully negative social experiences for autistic people:
[I learned] from an early age to think before I speak and to not be myself around people because I would always get into trouble for what I thought was just answering questions, continuing discussions and trying to explain how I felt. (Pumpkin)
This sense of danger meant that some participants felt they had to either socially isolate themselves for their own safety: ‘To avoid hurt and misunderstanding and because it makes life less painful for me I limit social contact[.]’ (Owain).
Theme 4: The constant effort of compensating and masking
Hiding autistic social differences or compensating for specific communication difficulties required constant concentration and effort for participants. One participant described the concentration and effort this constant masking required as ‘hypervigilance in most contexts’ (The Laird). Others described it as akin to being a detective or a psychologist: ‘I sometimes feel like a mini-psychologist when I talk to people as I am constantly trying to work out what they mean, their motivations, what they are trying to conceal.’ (Barney) and ‘I am constantly reading facial expressions, tone of voice, body language, voice fluctuations and analysing every word they say.[…] I sometimes feel like a detective on Criminal Minds or NCIS.’ (Pumpkin).
Much of this demand came from the intense scripting they had to do before conversations with others: ‘Mental rehearsal for conversations in the near future is a must for me […] it is very draining and takes up a lot of head space.’ (The Laird). Participants also had large number of, sometimes, highly elaborate strategies participants had developed to manage specific social situations such as starting or ending conversations, preventing themselves monologuing, or changing topics during conversations (omitted here for length; see Appendix B in the supplemental material available at https://osf.io/9cgz7 for examples).
The level of effort was such that participants likened communicating with neurotypical people as an autistic person to speaking a different native language to the rest of society, and therefore constantly having to translate:
I am really not sure any of my conversation skills are natural. It all feels thought through and performed. In some ways it is like translation software running in my brain. The person speaking uses words which I hear, but I than have to translate those words into my ‘language’, work out what I want to say back in my language, then translate that into their language, then try and say those words in the right order and with the correct accent! (Barney)
Theme 5: Finding and creating shared understandings across neurotypes
However, these social communication difficulties were not inevitable. Social interactions with other autistic people (and, to a lesser extent, other neurodivergent people in general) were easier, less stressful, and often preferable for participants. This was in part due to the shared expectations and values, which meant that participants did not have to monitor themselves (for monologuing, tangents, etc.) or others (for hidden agendas or non-literal language) so much: ‘I do enjoy the freedom of speaking frankly within my family (both my teenagers are autistic also) without all the silly window dressing’ (AppleFrog).
Shared lived experience and the opportunity to compare experiences with others was also important: ‘I have found peer support from other autistic people really helpful and learning what is ‘normal’ for my tribe has been really helpful.’ (Barney) Most important of all, however, was the fact that other neurodivergent people were often more understanding and accommodating than neurotypical people were:
I find the expectations of my Autistic friends to be quite different from my neuro-normative friends, for example I can set boundaries more effectively, I can say how I feel without it being cast as ‘rude’ and if I need to leave early they are okay with this. (Lily)
This meant, though, that neurotypical people who understood autism – and were willing to be flexible with their own social norms and values to accommodate participants – were also easier to be around, and desirable as friends:
I love my non autistic friends and now that we all know I’m autistic they don’t mind when I accidentally interrupt and I feel like I can relax and play with my hair or whatever when I’m talking to them and that I don’t have to make eye contact if I don’t want to, and in fact if I’ve had a hard day and I’m a bit shut down they will sort of scaffold group conversations around me without constantly asking me questions and trying to pull me back in when I can’t cope with it. (AppleFrog)
But because there was a disjunction in social expectations and lived experience between neurotypical people and participants (in a way there was not between neurodivergent people and participants), participants felt that knowing they were autistic was crucial. This let them understand how their social needs and expectations might differ to non-autistic people’s and to arrange their social interactions accordingly: ‘Becoming conscious or aware of one’s own shortcomings and creating a life compatible to one’s own operating system is key to have a contented life. For me it is a continues [sic] process in re-structuring my life accordingly.’ (The Laird).
4. Discussion
In this study, we aimed to elicit autistic perspectives on autistic social communication, and to identify features of a possible autistic social communicative style that has so far only been hinted at in other research. Specifically, we were interested in autistic perspectives on paralinguistic features of communication; differences communicating with autistic vs. non-autistic people; and other aspects of the autistic social communicative experience that participants felt researchers should know.
Participants perceived that they felt less strongly about the importance of paralinguistic phenomena than non-autistic people did. Some quotes by participants, especially from Theme 1, suggest that autistic people may simply use fewer and/or more idiosyncratic paralinguistic markers, an idea supported by extant literature (e.g. Rifai et al. 2022). It may be that paralinguistic markers are simply less relevant/necessary to the autistic interactional style than to the non-autistic one. Eye contact and hand gestures were singled out as features non-autistic people demanded in communication, but that participants felt were meaningless – again, supported by extant literature on reduced eye contact (Hessels et al. 2018) and different use of gesture (de Marchena and Eigsti 2010; de Marchena et al. 2019) by autistic people. Beyond paralinguistic phenomena, pragmatic features of language like euphemisms and sarcasm were actively considered hindrances to communication. Again, this is partially in line with extant findings that autistic people may find pragmatic language difficult (Pijnacker et al. 2009; Ying Sng, Carter, and Stephenson 2018; Vulchanova et al. 2015) – however, the framing of pragmatic language usage specifically as a hindrance to communication, rather than something participants found personally challenging, is novel.
More broadly, participants took a very permissive approach to social communication. There was an emphasis on clearly negotiated, individualised communication features between two people, personalised to that dyad’s relationship dynamic and preferences (as seen in Theme 2c: Mutual respect and effort) – again, a novel finding, though there are some similarities with Heasman and Gillespie’s (2019) findings. However, these individualised communication features did not seem to serve the same function as paralinguistic communication features do for non-autistic people. Instead, they seemed (e.g. Bert’s quote in 2c) to be closer to a very personalised version of the kinds of social scripts that non-autistic people use for greetings or other small talk (Methot et al. 2021).
Participants also felt there was a difference in social communication with autistic and non-autistic people. Some of those differences did seem fundamentally linked to a shared neurotype effect. For example, participants identified other autistic people as sharing social norms with them, or having shared lived experience, or being able to help them understand what was ‘normal’ for their neurotype. This is in line with claims made by the double empathy problem (Milton 2012; Milton 2014), and with other qualitative accounts of autistic-autistic interaction (Watts et al. 2024). However, other differences – like empathy, mutual accommodation, feelings of safety and acceptance – were not inherently neurotype-related. Instead, they were reflections of participants’ experiences with bullying, ableism, and enforced neurotypical norms in autism-hostile and predominantly neurotypical social environments; and ways in which being around other autistic people removed these threats. This is, again, in line with recent research that has begun to uncover how difficult navigating a neurotypically-normed world is for autistic people (Botha, Dibb and Frost 2022; Dekker 1999; Edwards et al. 2023; Kelly, O’Malley and Antonijevic 2018; Sinclair 2010), and the ways in which neurotypical people are (implicitly or explicitly) prejudiced against autistic people (Alkhaldi et al. 2021; Dickter et al. 2020; John, Knott and Harvey 2018; Sasson et al. 2017).
Finally, participants were eager to share many aspects of their social communication experiences that we had not anticipated – highlight the breadth of the autistic social communicative experience, but also the importance of researchers explicitly asking participants whether they are asking the right questions. These unanticipated aspects included positive experiences and things that brought them joy, the role physical and sensory environments played in their capacity to socialise and communicate – something that is only just beginning to be explored in the literature (e.g. Crompton, Hallett et al. 2020; Haydock et al. 2024; Watts et al. 2024). It also included the degree to which normalised ableism (and the related fear of interpersonal victimisation) impacted their social behaviour and communication with others. Autistic accounts of ableism as social/communicatively disabling are also only just beginning to be explored academically (e.g. Andoni 2024; Eliassen 2025; Nicolaidis et al. 2015; Roberts 2024), but are a well-established topic of discourse within autistic communities (Sinclair 1992, 294–302; Sinclair 1993; Sinclair 2010).
More broadly, regarding our guiding interest in autistic social communication and the autistic social communicative style, we did find some featural differences of autistic social communication. Some of these were more technical features: interrupting, monologuing (especially relating to special interests) and a ‘distractible’ (i.e. tangent-oriented) conversational style. All three of these are attested in the literature (e.g. Abbot-Smith et al. 2024; Williams, Wharton and Jagoe 2021; Ying Sng, Carter and Stephenson 2018). However, they are of specific interest here because some participants identified them as inherent social/communicative difficulties they experienced, which is more in line with traditional deficit framings of autism. This was a contrast to participants otherwise framing difficulties as a cross-neurotype relational or environment-related issue, which is more in line with newer neurodiversity framings of autism.
Most features, though, were more akin to a difference in social communication values: interacting for a specific purpose and talking about topics of interest; a reduced or easily depleted ‘social battery’; being honest, direct, clear, and authentic; and respecting other’s differences by putting the effort in to develop shared, negotiated interactional patterns in a relationship. In this, our findings echo Heasman and Gillespie’s (2019) and Williams et al.’s (2021) descriptions of broad differences in assumptions, expectations, relational experience, and purpose between autistic and non-autistic individuals during social interactions. Whilst these broader differences are valuable insights into the autistic social communicative style, they are notably different from the specific behavioural, nonverbal, and paralinguistic markers we were aiming to elicit. It is not fully clear why we were not able to identify these markers from the data collected. There are a few possibilities, which we mention the most likely two of here:
Firstly, some participant responses hint at these types of features being either reduced or absent from the autistic repertoire, which is in line with both the diagnostic criteria for autism (APA 2022; WHO 2022) and the wider literature on nonverbal communicative behaviour in autism (e.g. Rifai et al. 2022; Tager-Flusberg et al. 2005). It may be that paralinguistic markers are simply less relevant/necessary to the autistic social style than to nonautistic one.
Secondly, the difficulties some participants reported in maintaining awareness of their own behaviour during social interactions, or in ascertaining whether specific behaviours were natural or unnatural, may also have contributed. There is some evidence that autistic people may find explicit metacognition (about social communication, and also more generally) particularly difficult (Huggins et al. 2021; Morin, El-Sayed and Racy 2015). However, non-autistic people also find this difficult (Schön 2017; Frith 2012), especially in in – versus out-group contexts (Sieck 2017, 1–9), and often give very inaccurate reports of their own intentions and behaviour (Frith 2012). It therefore is possible that our choice of methodology (and/or the wording of our questions) is simply not suitable for eliciting this sort of information, from either autistic or non-autistic people. Indeed, much of our knowledge of the features of non-autistic social communication come from a) quantitative experimental evidence, b) observational studies, or c) self-reflection by academics who have been trained in how to do this. Such research on autistic social communicative features, by comparison, is in its infancy – especially given much of autism research has traditionally focused on judging autistic social communication by non-autistic standards and norms, rather than characterising it as its own phenomenon.
4.1 Limitations
Participants were all adults from the UK and were primarily white. No participants disclosed having a co-occurring intellectual disability or being non-speaking when asked about additional conditions. It is therefore unclear whether our findings generalise to the wider autistic population. Further work to is needed to ascertain how the themes identified here do or do not apply to autistic people with an intellectual disability, who are non-speaking, who live in countries with different social norms to the UK, or who are children.
4.2 Future directions
This study provides valuable insight into autistic people’s social communication experiences and values, but the differences it highlights between autistic and non-autistic people are entirely based on the autistic participants’ suppositions about what the non-autistic social communicative experience is like. Whilst this is of interest in its own right (especially given claims of Theory of Mind deficits in autism (Baron-Cohen 2000), which would seem to preclude the kinds of social perspective-taking demonstrated here), it has the same pitfall as historical autism research does, but in reverse: one group of people, on one side of a mutual relational disjunction caused by lack of shared insight into each other’s experiences (Milton 2012), is attempting to identify where the disjunction lies by speculating about the other group’s experiences.
Some previous studies have attempted to address this issue, for example by using shared reading practices between autistic and non-autistic people to encourage mutual understanding and relation (Chapple et al. 2021; Chapple et al. 2022). A similar comparative, co-operative approach on the topic of social values might provide better insight into where expectations and values truly diverge between neurotypes, whilst also uncovering similarities between autistic and non-autistic social communication experiences – potential points of reference with which to bridge the double empathy gap.
Nonetheless, this study’s holistic framing of conflicts between autistic and non-autistic social communicative styles as a difference in values and expectations is both novel and useful. It suggests ways of shifting away from the outdated ‘deficit’ framings common in autism research, without falling into the trap of attempting to identify universal autistic ‘strengths’ to outweigh these. Future work in this area should explore this concept further, with a focus on understanding how different neurotypes impact people’s expectations about the function of social communication and how interactions should proceed.
Data Accessibility Statement
Data associated with this study are not openly available.
Notes
Ethics and Consent
Ethical approval was granted by the University of Edinburgh Medical Research Ethics Committee. All participants provided informed consent before starting.
Acknowledgements
We would like to thank the autistic people who contributed their time, experiences, and insight as part of this study – Zel, AppleFrog, Pumpkin, The Laird, Barney, Owain, Bert, Lily, and Alexander. We would also like to thank everyone who provided feedback on this research at different stages of the analysis, and at the various conferences and meetings that this research was presented at.
Competing Interests
The authors have no competing interests to declare.
Authors Contributions
Holly E.A. Sutherland: Conceptualisation, methodology, validation, formal analysis, investigation, data curation, writing – original draft, writing – review & editing, visualisation, project administration, funding acquisition. Catherine J. Crompton: Conceptualisation, methodology, writing – review & editing, supervision, project administration, funding acquisition. Joseph Long: Conceptualisation, methodology, writing – review & editing, supervision, project administration, funding acquisition. Sue Fletcher-Watson: Conceptualisation, methodology, writing – review & editing, supervision, project administration, funding acquisition.